Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 1
AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT USING ANIMATED DEMONSTRATIONS
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 2
Abstract
The purpose of this study was to consider animated demonstrations in the context of cognitive load theory and to
determine if demonstration users would exhibit a delayed performance decrement, described as Palmiter’s animation
deficit. Relative condition efficiency (RCE) was measured, but in addition a new measure, Performance efficiency
(PE) was developed. Given the worked example effect, it was hypothesized that the demonstration learners would
out-perform those in a practice condition. Results found the animated demonstration groups assembled the week one
problem in significantly less time than the discovery-practice group, providing positive evidence for the worked
example effect. Significant results were also found for relative condition efficiency and performance efficiency.
Thus the performances of the demonstration groups were found to be significantly more efficient than those of the
discovery-practice group. Also, group performances did not differ a week later, providing no evidence of an
animation deficit.
Keywords: animation, cognitive load theory, performance efficiency
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 3
AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT USING ANIMATED DEMONSTRATIONS
Abstract The purpose of this study was to consider animated demonstrations in the context of cognitive load theory
and to determine if demonstration users would exhibit a delayed performance decrement, described as Palmiter’s
animation deficit. Relative condition efficiency (RCE) was measured, but in addition a new measure, Performance
efficiency (PE) was developed. Given the worked example effect, it was hypothesized that the demonstration
learners would out-perform those in a practice condition. Results found the animated demonstration groups
assembled the week one problem in significantly less time than the discovery-practice group, providing positive
evidence for the worked example effect. Significant results were also found for relative condition efficiency and
performance efficiency. Thus the performances of the demonstration groups were found to be significantly more
efficient than those of the discovery-practice group. Also, group performances did not differ a week later, providing
no evidence of an animation deficit.
Keywords: animation, cognitive load theory, performance efficiency
Introduction
Sweller and Cooper (1985) compared the performance of those who had learned by studying worked
examples to those who learned while solving problems and found that those who studied worked examples had a
decreased solution time and a decrease in the number of mathematical errors (Cooper & Sweller, 1987; Sweller &
Cooper, 1985). This was later described as “the worked example effect” (Sweller & Chandler, 1991). While the
worked example effect has been replicated many times and is the most studied cognitive load effect (Sweller, 2006),
there are limitations to this effect. As learners become more competent, the worked example and other cognitive
load effects begin to disappear (Kalyuga, Chandler, & Sweller, 1998). This change in learner behavior was termed
the “expertise reversal effect,” and researchers found that those with more expertise benefited more by solving
problems (Kalyuga, Ayres, Chandler, & Sweller, 2003).
Educational researchers have been recommending that novices study worked examples during early schema
acquisition (Jonassen, 1997; Sweller, 1988; Sweller & Chandler, 1991) and have even argued that novices study
worked examples, rather than learn through discovery problem solving (Sweller, 1988; Touvinen & Sweller, 1999).
Chandler and Sweller (1992) defined worked examples as “a problem statement and the appropriate steps to
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 4
solution” (p. 233). More recently, Sweller and his colleagues were much more explicit and defined worked examples
as “a step-by-step demonstration of how to perform a task, or how to solve a problem” (Clark, Nguyen, Sweller,
2006, p. 190). This paper proposes this definition may apply to animated worked examples, or as they are described
in the literature, animated demonstrations (Lipps, Trafton, & Gray, 1998; Palmiter, 1991; Waterson & O’Malley,
1993). Yet, the literature concerning animated demonstrations often cites Palmiter’s early studies, which describe
animated demonstration groups as having a decreased performance in a delayed test (Bétrancourt & Tversky, 2000;
Palmiter, 1993; Tversky, Morrison, & Betancourt, 2002). This effect was also described as an “animation deficit”
(Lipps et al., 1998, p.1). Therefore, if animated demonstrations act as animated worked examples, this “animation
deficit” is in direct conflict with Sweller and Chandler’s “worked example effect.”
While the subject matter of cognitive load studies was originally confined to science and mathematics,
researchers eventually considered other types of problem solving and found that the worked example effect extends
to other domains [e.g. music, chess, athletics (Atkinson, Derry, Renkl, & Wortham, 2000)]. This effect was even
recently extended to ill-structured domains like art and design (Rourke & Sweller, 2009). The current study
analyzed the performance of novices as they learned how to use a graphics editing program. Like Touvinen and
Sweller (1999) and Palmiter (1991), this study compares several instructional strategies (and combinations of these
strategies) to monitor novice behavior during procedural learning.
Modality and cognitive load
In the late 1990s, Mayer and his colleagues proposed the “cognitive theory of multimedia learning”
(Mayer, 1997; Mayer, 2001; Mayer & Moreno, 1998), which relies on cognitive load theory, but also draws from
dual-coding theory and constructivism (Mayer & Moreno, 2002). According to the cognitive theory of multimedia
learning, humans process visual and auditory information during multimedia instruction; have a limited working
memory; and learn by selecting, organizing and integrating information from the environment into their long term
memories (Mayer, 2001).
Mayer and his colleagues have been quite prolific over the past two decades, and are well known for their
contributions to the “modality effect” or “modality principle” (Mayer, 2001; Mayer, 2005; Moreno and Mayer,
1999). These researchers have conducted many empirical studies of animated instruction. A common theme among
these studies is that the instructional materials revolved around the learner’s application of principles portrayed in
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 5
animated instruction of causal systems. In each case, learners were shown animated representations of how a
dynamic system works. These systems included hydraulic braking systems (Mayer, Deleeuw, & Ayers, 2007), a
bicycle pump (Mayer & Anderson, 1991; Mayer & Sims, 1994), and lightening formation (Mayer, Bove, Byman,
Mars, & Tapangco, 1996; Mayer & Moreno, 1998) [A more thorough explanation of these studies is explained in
Mayer (2009)]. In each case, learners were required to answer retention and transfer questions. The required
performance was to respond to verbal statements and provide written responses about causal systems as opposed to
how to perform a task, or how to solve a problem. Since the development of Mayer’s cognitive theory of multimedia
learning, many studies have followed Mayer’s lead to consider animated instruction. However, learning extends
beyond an understanding of causal systems.
Multiple forms of learning by animation
Researchers from various fields have explored human behavior to develop a biological model of learning,
based on human anatomy, physiology, and genetics (Anderson, Bothell, Byrne, Douglass, Lebiere, & Qin, 2004;
Squire, 2008; Clark & Clark, 2010). By the early 1990s, the physiological literature had defined two broad classes of
learning as either procedural or declarative (Squire, 1992). Declarative learning is concerned with the learning of
language-based information (e.g. facts and events), while procedural learning is skills-based learning (e.g. learning
how to use a computer program) (Squire & Zola, 1996). While this distinction may seem to be an oversimplification
from an educator’s perspective, it is based on several decades of physiological research that dealt with the learning
capabilities of brain injured patients, primates, and normal humans (Squire, 2008; Scoville & Milner, 1957).
If one were to consider the educational literature for a nomenclature of learning, they might choose Clark
and Mayer (2008) who describe five types of learning (facts, concepts, procedures, processes, and principles). The
literature describing these five types of learning also extends back several decades and relies on many contributors,
most notably Merrill (1983). The nomenclatures of Squire and Zola (1996) and Clark and Mayer (2008) are
compatible, and there is quite a bit of overlap between them. The underlying message of this discussion is that
educators should not think of human learning or memory as a singular entity (Merrill, 1983), but be specific about
what type of learning they are trying to encourage. In doing so, it will help to develop a productive set of theories
and contribute to the development of the learning sciences.
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 6
Niajar (1996) made a distinction between two different types of animated instruction. He describes
animated instruction as being either “animated explanation,” or “animated demonstration.” Mayer’s instructional
materials are more aptly described as animated explanation (Niajar, 1996). Mayer describes his work well when he
says his experiments ask questions about scientific explanation: “By ‘explanation’ we mean a description of a causal
system containing parts that interact in a coherent way, such as a description of how a pump works or how the
human respiratory system works” (Mayer & Sims, 1994, p.389). Clark and Mayer (2008) even describe “two
different e-learning goals,” which teach learners to “inform and perform” (p.17). We agree with this distinction and
terminology to propose learning “how to do” something (perform) can be taught via animated demonstration, as
opposed to learning about something (inform), which is taught via animated explanation.
Articles from outside the instructional technology or educational psychology literature, also describe these
forms of learning as being quite different. John Anderson’s ACT-R framework, for instance, provides a well-
accepted theoretical basis for considering explanation and demonstration as two different forms of learning
(Anderson, 1993), requiring different forms of instruction. While Anderson’s original model considered a
declarative only origin of procedural skill acquisition (Anderson, 1976; Anderson, 1983), his later work found that
skill acquisition may also develop from example-based processing, to eventually become rule-based processing
(Anderson & Finchman, 1994; Anderson, Fincham, & Douglass, 1997). Thus examples may be initially represented
in working memory, and then later, after practice, generalized into production rules (Anderson et al., 1997).
This study follows on the above literature to propose a dichotomy of animated instructional materials based
on the learning involved, as being either declarative or procedural. Mayer’s animated explanations (declarative)
require learners to encode information about how a system works. Animated demonstrations (procedural) are
similar, but learners must consider rules based upon a sequence of actions, which they eventually must perform.
Thus animated demonstrations may be an important, cognitively demanding form of instruction, providing a fertile
area for cognitive load research.
Animated instruction for procedural learning
For the purposes of this study, an animated demonstration is defined as a narrated presentation that shows
learners how to perform a computer-based procedure. Palmiter’s dissertation project is perhaps the most cited
animated demonstration study (Palmiter, 1991). In a series of experiments, Palmiter compared the learner
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 7
performance of those who studied narrated animated demonstrations with those who used text-based instruction
(Palmiter & Elkerton, 1991; Palmiter, 1993). Palmiter’s results are quite interesting, for she found that during a
training session learners who studied animated demonstrations sped skill acquisition, and performed tasks in less
time and more accurately than their peers using text-based instruction (Palmiter, 1991; Palmiter & Elkerton, 1991).
These results are similar to those by Sweller and Cooper (1985) who demonstrated that studying worked examples
requires “considerably less time to process than conventional problems, but that subsequent problems similar to the
initial ones also were solved more rapidly”(p.59).
However, Palmiter noted that, one week later, learners using animated demonstrations took longer to
perform tasks and reported that even though performance may have been quicker during the initial training session,
their retention was lacking a week later. This delayed performance decrement was later described as an animation
deficit (Lipps et al., 1998). However there is reason to question this decrement, since several other researchers have
been unable to replicate Palmiter’s findings (e.g. Cornett, 1993; Harrison, 1995; Lipps et al., 1998; Reimann &
Neubert, 2000; Waterson & O’Malley, 1993).
Considerations for this study
Since Sweller (1988) published his initial article outlining cognitive load theory, constructivism has
garnered a strong following. Constructivists soon began to promote the idea of activity as a means of learning with
technology (Brown, Collins, & Duguid, 1989; Jonassen & Rohrer-Murphy, 1999). While it has long been known
that activity can reinforce learning (e.g. Skinner, 1958), Sweller and his associates questioned the timing of that
activity during early schema acquisition. Tuovinen and Sweller (1999) found evidence that worked example-based
instruction was more beneficial for novices than discovery learning. Even though there is considerable evidence for
the worked example effect (Sweller, 2006), some question this instructional technique and promote the idea of
withholding assistance or guidance from learners (Koedinger & Aleven, 2007).
Given Sweller and Cooper’s early results, and Mayer’s work with declarative multimedia, it was
hypothesized that a narrated, animated, worked example would promote procedural learning. This research should
be testable and falsifiable for there is literature describing both a positive and negative effect given animated worked
examples. The overall research question of the study is: How will animated demonstrations impact student learning?
In order to answer this general question a number of key variables were examined. Sweller and Cooper’s early
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 8
worked example studies found differences in other outcome variables like completion times and the number of
errors (Sweller & Cooper, 1985). The animated demonstration literature considers similar outcome variables, but
followed Palmiter’s lead, to describe these variables as performance time and accuracy (Waterson & O’Malley,
1993; Lipps et al., 1998; Palmiter & Elkerton, 1991). Therefore this study used these variables to consider the
worked example effect given animated demonstrations.
Method
Participants
The participants of this study were undergraduate pre-service teachers (N=122) enrolled in an introductory
instructional technology course at a large university in the southeastern United States. The mean age of this sample
was 20.2 years (SD=2.89), and they were primarily female (93 females and 29 males). There were 14 freshmen, 60
sophomores, 42 juniors, and 6 seniors. Participants were randomly assigned to one of four instructional conditions,
demo (n=33), demo+practice (n=29), demo2+practice (n=36), practice (n=24). An a priori power analysis for a four
group MANOVA produced a sample size of n=115 participants. This number of participants is necessary to arrive a
power of 0.80, with a small effect size η2= 0.125, given α =0.05 (α=0.05 is used throughout this article, unless stated
otherwise) (Stevens, 2002).
Materials
While the worked example effect is well documented, Koedinger and Aleven (2007) proposed this effect
may be an artifact of poor control conditions. Schwonke, Renkl, Krieg, Wittwer, Aleven, & Salden (2009)
considered this proposal, and found evidence that the worked example effect held, even when problem solving was
well-supported by feedback and scaffolding. Therefore it is suggested that those considering the worked example
effect present all learners with an identical instructional overview of the topic being considered, so that all learners
are properly prepared.
Paas (1992) and Tuovinen & Sweller (1999) provided participants with an introductory overview of the
subject matter. The current study continued this practice to develop an introductory overview. This overview was a
narrated, non-animated, PowerPoint presentation (Microsoft, 2003) that was encoded for web presentation with
TechSmith Camtasia Studio (TechSmith, 2006). It provided learners with an introduction to digital image editing
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 9
with Adobe Photoshop Elements, by presenting learners with a narrated presentation of screenshots. Specifically,
learners were shown how to select, move, rotate, and hide layers within an Adobe Photoshop Elements document.
Thus all learners were shown the basic procedural steps required to complete the practice problem.
Two animated demonstrations were developed with TechSmith Camtasia (TechSmith, 2006). The first
demonstration was identical to the week one practice problem (the mimicry condition). A second (varied context
condition) used an animated demonstration that showed the construction of a photo collage document. Both were 10
minutes long and demonstrated the same underlying procedures, but within different contexts. The animated
demonstrations showed learners how to select, move, rotate, and hide layers.
Two problem scenarios (the Mr. Potato head problem & the Picnic problem) were developed for this study.
Each was an assembly task that required assembly within Adobe Photoshop Elements (Adobe Systems, 2002). The
performance objectives of each problem required learners to select, move, rotate, and flip layers to produce a final
product. During week one, three groups of learners put together the Mr. Potato head problem. One of the goals of
this project was to measure what learners would retain and apply in a delayed performance. So, one week after
initial instruction (week two), learners put together another Adobe Photoshop Elements document (the Picnic
problem) that required learners to recall what they had learned one week prior.
Procedure
Before learners entered the test environment, TechSmith Morae Recorder (TechSmith, 2004) was turned on
and was allowed to record all onscreen actions as learners interacted with the instruments and problem-solving
scenarios. In addition, an image of the problem scenario was projected on a large screen in the room throughout
problem construction.
Learners began by first completing an initial demographics survey. The action of submitting this survey
launched the instructional overview. When the overview concluded, a JavaScript program randomly divided learners
into one of four instructional conditions:
demo - learners reviewed an animated demonstration, but did not practice with the “Mr. Potato
head” problem;
demo + practice - learners reviewed the identical animated demonstration, then practiced their
newly learned skills with the “Mr. Potato head” problem;
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 10
demo2 + practice - learners reviewed a different, collage-based animated demonstration and then
practiced with the “Mr. Potato head” problem;
practice - learners practiced with the “Mr. Potato head” problem, but did not review an animated
demonstration.
Following the instructional conditions, learners were asked to complete a post-treatment survey. This
survey included a relative condition efficiency question, “I invested:” with nine possible responses, from “very, very
low mental effort” to “very, very high mental effort” as described by Paas and van Merriënboer (1993). Once
learners had finished the post-treatment survey they were thanked for their participation, asked not to discuss the
instruction and not to use the software before the delayed test session (one week later). Once learners had left the
room, a researcher visited the learner’s station and saved the recording for later processing.
A week later, during the delayed test session, all groups of learners solved the picnic problem. Again, an
image of the completed product was projected on a large screen in the room throughout problem construction and
learner on-screen actions were recorded. After attempting the picnic problem, learners took a second post-treatment
survey, which also included a relative condition efficiency question.
Calculation
Paas and van Merriënboer (1993) derived the “relative condition efficiency” construct, which is composed
of standardized mental effort ratings and performance scores (see Equation 1) and allows researchers to compare
instructional conditions (Paas & Van Merriënboer, 1993). Learners rate their instructional conditions based upon
their “perceived” mental effort during instruction. Later their performance scores and “perceived” mental effort
ratings are combined in the relative condition efficiency construct (see Equation 1), and are presented graphically
(see Figure 1).
Like Paas and van Merriënboer (1993), this study develops a new construct called performance efficiency
(PE) (see Equation 2). Performance efficiency was developed to complement relative condition efficiency (RCE),
and is used to study human behavior given instructional conditions. Performance efficiency is similar to relative
condition efficiency (see Equation 2), but replaces the perceived mental effort rating with a more objective measure,
performance time (compare Equations 1 and 2).
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 11
RCE 2
rtMentalEffoZePerformancZ (1)
PE 2
eTimePerformancZePerformancZ (2)
Since performance efficiency does not include a mental effort rating, it is not a measure of cognitive load;
however like RCE, PE may also be used to describe group performances, relative to different instructional
conditions but in a more objective manner. Performance efficiency contrasts instructional conditions in much the
same manner, but combines two performance variables in a biplot, with group performance times and performance
scores on the x and y axes (respectively). In keeping with the animated demonstration literature, performance in the
above equations was described as accuracy.
Measurement of the dependent variables
This research was based in part, on a cognitive model from Human-Computer Interaction (HCI) research,
the GOMS model (Card, Moran, & Newell, 1983). The GOMS model analyzes computer-based problems by
reducing them to solution steps, known as problem solving operators (Dix, Finalay, Abowd, & Beale, 2003).
Accuracy, one of the dependent variables, was measured by using a rubric or checklist, based upon the problem-
solving operators of the problem being solved. Rubrics were developed, based on the problem scenarios, and
constructed to automatically total the points scored in an Excel spreadsheet (see Table 1) (Microsoft, 2003). A
researcher reviewed the Morae recordings of learner onscreen actions, and logged learner performances scores
according to the rubric. Learners were either given credit for completing a problem solving operator or not.
Successful completion of item movement or rotation was scored one point, and successful completion of layer
placement and flipping was given two points.
In addition, performance time was measured in seconds with TechSmith Morae. This software allows a
researcher to mark the timeline of a learner’s onscreen recording and to designate the beginning of a performance, in
this case with the first initial mouse movement. The end of the performance was also marked, and was designated as
the point when the greatest number of problem solving operators had been completed.
Results
A MANOVA was used to analyze the two outcome variables, the performance time and accuracy of each
performance. The assumptions of a MANOVA were analyzed for those learners who followed the instructions,
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 12
completed all surveys, and attempted the required performances; this group was considered the sample (N=122).
According to Glass and Hopkins (1984), these data met the independence assumption since the observations were
independent of one another. A SAS macro %MULTNORM (SAS, 2007) revealed non-normality (violating the
normality assumption), when the Shapiro-Wilks’ W= 0.76 p<0.0001 for accuracy, and for performance time the W =
0.95, p=0.0015. Mardia skewness was found to be β1,p= 146.5, p<0.0001 and kurtosis was β2,p=12.01, p<0.0001
(Mardia, 1970). This violation was primarily due to a set of multivariate outliers (n=34), which were subsequently
removed, by using the %MULTNORM macro. Next the data were transformed to test the “homoscedasticity”
assumption. Box’s M test (Box, 1954) was performed and X2 (3, N=88) =4.50, p=0.21, φ=0.23, therefore, the
variance-covariance matrices were not found to differ significantly, so there was no evidence that the
homoscedasticity assumption was violated.
Recall that the demo group (n=19) did not practice during week one, so the performance of practicing
learners (n=69) was analyzed. Group composition was demo+practice group (n=21), demo2+practice group (n=31),
and practice group (n=17). Week one group performances (performance time and accuracy) were found to be
significantly different, since Wilks’ Λ = 0.68, F (2, 68) = 6.83, p <0.0001, η2=0.32. Post hoc comparisons with
Scheffé’s test (p<0.025) revealed that both demonstration groups (demo+practice and demo2+practice groups)
assembled the problem in significantly less time than the practice group.
The general procedure for analyzing relative condition efficiency (Paas & van Merriënboer, 1993) was used
to compare groups. Group relative condition efficiency scores were compared with an ANOVA. Because the demo
group did not have a week one performance, the performances of three practicing groups were compared. The
assumptions of this ANOVA were analyzed and the data set met these assumptions, thus it was reasonable to
consider the ANOVA. The ANOVA was conducted and revealed that there were significant differences between
groups since F (2, 68) = 3.69, p=0.03.
Week one performance efficiency (see Equation 2) was also found to be significant, since F (2, 68) = 12.95,
p<0.0001. Post hoc comparisons with Scheffé’s test (p<0.05) revealed significant differences between groups, with
the demonstration groups (demo+practice and demo2+practice) having more efficient performances than the
practice group, given performance efficiency.
The week two performance was analyzed with a MANOVA to determine group differences a week after
initial instruction. The results of the MANOVA found that there was not a significant difference given learner
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 13
performance one week after initial instruction, since Wilks’ Λ =0.96, F (3, 87) =0.64, p =0.70, η2=0.04. During
week two, relative condition efficiency was also considered, by combining the week two accuracy scores with the
week two perceived mental effort ratings. Group z-scores were tested with an ANOVA and revealed that there were
no significant differences between groups as F (3, 87) = 0.38, p=0.77. Week two performance efficiency was also
calculated in a similar manner. An ANOVA for week two performance efficiency revealed no significant differences
in group means, because F (3, 87) = 0.42, p=0.74.
Discussion
Learners who studied animated demonstrations assembled the week one problem in significantly less time
than those who learned through discovery problem solving. This result is consistent with our initial hypothesis, the
worked example effect and previous research (Sweller & Chandler, 1991; Tuovinen & Sweller, 1999), and it is
further evidence of the worked example effect, given animated demonstrations. As expected, those learners who
were provided with direct methods of instruction, in this case animated demonstrations, improved performance
efficiency during early schema acquisition. Those who studied an animated demonstration, that was identical to the
problem scenario (the demo+practice or mimic condition), had a lower performance time and increased accuracy
over those in the practice group. In addition, those learners who studied a different animated demonstration (varied-
context) significantly outperformed those who only learned through problem solving. Therefore, this study
established animated demonstrations can indeed act as animated worked examples, and that they are an effective and
efficient form of instruction. Instructional designers at companies like Bank of America, Amazon, and Microsoft,
are all using narrated web-based animated demonstrations (demos) to train novices to use their online products and
services. Therefore, it is the recommendation of this study that these developers continue to use this effective and
now, evidence-based, e-learning strategy.
Group performances did not differ on the delayed test. Thus the demonstration learners retained their skills,
and performed equally well one week after initial instruction. This finding does not support Palmiter’s animation
deficit, and is consistent with the results of other researchers (Lipps et al., 1998; Waterson & O’Malley, 1993).
Tuovinen and Sweller (1999) proposed retention may be an issue given worked examples, and asked future
researchers to consider the durability of learning given worked example-based instruction. This result provides some
evidence for the durability of worked example-based instruction. Other researchers should also consider Tuovinen
and Sweller’s proposal.
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 14
However it should be clearly stated that the animated demonstrations used in this study were not like those
of Palmiter’s study, for they were narrated. Narration is heavily suggested given the modality and split-attention
effects (Chandler, & Sweller, 1992; Moreno & Mayer, 1999; Penney, 1989; Mousavi et al., 1995). The addition of
narration to an animated sequence is beneficial to learners and is described as the “modality effect” (Mayer, 2001).
In an asynchronous e-learning environment, an instructor’s role may be compromised because of an inability to
communicate with an e-learner, but the use of a narrated animated demonstration allows that instructor to overcome
that barrier to provide e-learners with guidance, support and “just-in-time” training. Palmiter may have been correct,
asserting that only providing learners with animation produces mimicry, what Ausubel may have described as rote
learning (Ausubel, 1963). The importance of adding narration to an animated demonstration should not be
underestimated, for it promotes what Mayer (2001) described as “multimedia learning,” now even with procedure-
based instruction. Narrated animated demonstrations provide learners with a verbal narrative, which directs a
learner’s attention during the presentation, and more importantly, provides the learner with an expert level
explanation.
Palmiter chose to study students who learned individual discrete tasks. Each of Palmiter's articles describe
tasks (e.g. copy button or copy field) in isolation, and not in the context of an overall problem; whereas, the current
study investigated learners in situ. That is, learner performance was studied given an authentic context, as they used
their skills as a part of a larger project. This was necessary to gather data concerning a learner’s cognitive load
during actual problem solving, but also to measure learning in an ecologically valid manner.
Animated demonstrations present learners with a problem and a solution for solving that problem (the
underlying schema). Those who use animated demonstrations rarely find a problem solution that exactly matches the
problems they encounter, so they must extract a schema from an animated demonstration to later apply that schema
in a new context. The results from the varied context condition (demo2+practice) showed learners do indeed extract
that problem schema and improved their subsequent performance, to the point where their group performance was
significantly different from those who discovered a problem solution. Given these results, Palmiter’s critique of all
animated demonstrations is unfounded.
It should be reiterated that the worked example effect has its limitations. Recall that Kalyuga et al. (2003)
found evidence of an expertise reversal effect. In other words, those learners who had more experience with the
subject matter benefited less from worked example-based instruction. Learners with more expertise benefited more
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 15
from problem solving. This limitation has implications for animated demonstrations. Because of the expertise
reversal effect, it would be expected that animated demonstrations would prove to be redundant and less useful for
those with more expertise; perhaps this question can be answered by future researchers.
Relative condition efficiency and other similar metrics have become a basis for much of cognitive load
measurement in instructional technology (Paas, Tuovinen, Tabbers, & Van Gerven, 2003; Tuovinen & Paas, 2004),
but this measurement relies on indirect or subjective measures, therefore it is unclear how this construct relates to
actual cognitive load (Brünken, Plass, & Leutner, 2003). This study developed a new metric for measuring the
efficiency of learner performance given an instructional condition. Performance efficiency is similar to relative
condition efficiency, but does not include perceived mental effort in the equation. While perceived mental effort is
related to performance, performance efficiency focuses on the efficiency of a performance. There probably is a
relationship between the mental effort endured during an instructional condition and performance related outcome
variables, but performance efficiency does not consider that relationship. It is only a way to compare and relate the
performances of groups given instructional conditions. Performance efficiency is a useful metric, because it is
generalizable. It is conceivable that any performance measure could be contrasted with its performance time,
graphed and analyzed. Therefore, this metric may be used outside of the cognitive load literature to be considered by
other groups of researchers. However, it has its limitations and should only be applied to procedure-based
instructional conditions. It is hoped researchers will use both of these metrics to refine instructional materials and
improve learner performance.
The finding of no significant differences a week after initial instruction may seem to be a reason to discount
the need for animated worked examples. However the finding of improved performance efficiency during the initial
training period is reason to consider this presentation strategy. An efficient performance is important for it showed
this group of learners is functioning at a higher level, early in schema acquisition. This improved performance is
evidence that this group of learners is better prepared. Therefore it could be argued that they are better able to learn
additional procedures whereas, the practice group may very well be overloaded by additional training.
Another impact of this study is that the recording methodology may be generalized to any e-learning
environment. Researchers may record learner on-screen actions to document how novices behave and react when
tasked with an unfamiliar learning environment. In this study, recordings documented learner errors and the problem
solving operators that they employed. Thus educational researchers may use recording technologies to consider
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 16
learner performance, given various instructional strategies and environments. They may do so from a cognitive load
perspective, to contrast the relative condition efficiency of instructional alternatives. Now, they may do so from a
performance perspective, to contrast the “performance efficiency” of instructional strategies. It is hoped these tools
and methodologies will be used together to promote the learning sciences.
In conclusion, it has been proposed that learners will be more productive if they teach themselves how to
perform procedures (e.g. Bruner, 1961), but like Tuovinen and Sweller (1999), this study cannot support that
position. This position diminishes the role of an instructor or instructional designer. These educators have purpose.
As Kirschner, Sweller & Clark (2006) proposed, they provide guidance and support. Given the results of this study,
it is more ethical for an instructor to take action and guide learners during early schema acquisition, rather than
withhold that assistance to leave them to teach themselves.
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 17
References
Adobe Photoshop Elements (Version 2.0) [Computer program] Mountain View, CA: Adobe Systems
Anderson, J. R. (1976). Language, memory, and thought. Hillsdale, NJ: Erlbaum.
Anderson, J. R. (1983). The architecture of cognition. Cambridge, MA: Harvard University Press.
Anderson, J. R. (1993). Rules of the mind. Hillsdale, NJ: Erlbaum.
Anderson, J. R., Bothell, D., Byrne, M. D., Douglass, S., Lebiere, C., & Qin, Y . (2004). An integrated theory of the mind. Psychological Review 111 (4). 1036-1060. doi:10.1037/0033-295X.111.4.1036
Anderson, J.R., & Fincham J.M. (1994). Acquisition of procedural skills from examples. Journal of experimental psychology: Learning, memory, and cognition, 20 (6) 1322-1340. doi:10.1037/0278-7393.20.6.1322
Anderson, J.R., Fincham, J.M. & Douglass, S. (1997). The role of examples and rules in the acquisition of a cognitive skill. Journal of experimental psychology: Learning, memory, and cognition, 23(4) 932-945. doi:10.1037/0278-7393.23.4.932
Atkinson, R.C., & Shiffrin, R.M. (1968). Human memory: a proposed system and its control processes. In K.W. Spence (Ed.), The psychology of learning and motivation: Advances in research and theory, Vol. 2 (pp. 89–195). New York: Academic Press.
Atkinson, R. K., Derry, S. J., Renkl, A., & Wortham, D. W. (2000). Learning from examples: Instructional principles from the worked examples research. Review of Educational Research, 70, 181–214. doi:10.2307/1170661
Ausubel, D.P. (1963). The psychology of meaningful verbal learning; an introduction to school learning. New York: Grune and Stratton.
Bétrancourt, M., & Tversky, B. (2000). Effect of computer animation on users' performance: a review. Le travail Humain, 63(4) 311-330.
Box, G. E. P. (1954). Some theorems on quadratic forms applied in the study of analysis of variance problems. Annals of Mathematical Statistics, 25, 290-302. doi:10.1214/aoms/1177728717
Brown, J.S., Collins,A. & Duguid, P. (1989). Situated cognition and the culture of learning. Educational Researcher, 18 (1) 32-42. doi:10.2307/1176008
Bruner, J. S. (1961). The act of discovery. Harvard Educational Review, 31(1) 21–32.
Brünken, R., Plass, J. L., & Leutner, D. (2003). Direct measurement of cognitive load in multimedia learning, Educational Psychologist, 38(1), 53–61. doi:10.1207/S15326985EP3801_7
Card, S., Moran, T., & Newell, A. (1983). The psychology of human-computer interaction. Lawrence Erlbaum Associates, Hillsdale, NJ
Chandler, P., & Sweller, J. (1991). Cognitive load theory and the format of instruction. Cognition and Instruction, 8(4), 293-332. doi:10.1207/s1532690xci0804_2
Chandler, P., & Sweller, J. (1992). The split-attention effect as a factor in the design of instruction. British Journal of Educational Psychology, 62, 233-246.
Clark, R.E., & Clark, V. P. (2010). From Neo-Behaviorism to neuroscience: Perspectives on the origins and future contributions of cognitive load research. In J. Plass, R. Moreno, & R. Bruken (Eds.), Cognitive load: Theory and application. New York: Cambridge University Press.
Clark, R.C., & Mayer, R. E. (2008). E-learning and the science of instruction (2nd ed). San Francisco: Jossey-Bass.
Clark, R.C., Nguyen, F., & Sweller, J. (2006). Efficiency in learning: evidence-based guidelines to manage cognitive load. San Francisco: Pfeiffer.
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 18
Cooper, G., & Sweller, J. (1987). Effects of schema acquisition and rule automation on mathematical problem-solving transfer. Journal of Educational Psychology, 79(4), 347-362. doi:10.1037/0022-0663.79.4.347
Cornett, L. L. (1993). Animated demonstrations versus text: a comparison of training methods (Unpublished Master’s Thesis). Rice University, Houston, TX.
Dix, A., Finlay, J., Abowd, G. & Beale, R. (2003). Human-Computer Interaction. (3rd ed). Essex: Pearson Education Limited
Harrison, S. M. (1995). A comparison of still, animated, or nonillustrated on-line help with written or spoken instructions in a graphical user interface. In I. R. Katz, R. Mack, L. Marks, M. B. Rosson, & J. Nielsen (Eds.), Proceedings of the ACM conference on human factors in computing systems (pp. 82-89). Denver, CO: Association for Computing Machinery.
Jonassen, D.H. (1997). Instructional design models for well-structured and ill-structured problem-solving learning outcomes. Educational Technology Research and Development, 45(1), 65–94. doi: 10.1007/BF02299613
Jonassen, D.H & Rohrer-Murphy, L. (1999). Activity theory as a framework for designing constructivist learning environments. Educational Technology Research and Development, 47 (1) 61-79 doi: 10.1007/BF02299477
Kalyuga, S. Ayres, P., Chandler, P. & Sweller, J. (2003). The expertise reversal effect. Educational Psychologist, 38(1) 23–31. doi:10.1207/S15326985EP3801_4
Kalyuga, S., Chandler, P., & Sweller, J. (1998). Levels of expertise and instructional design. Human Factors, 40 (1) 1-17. doi:10.1518/001872098779480587
Kirschner, P. A., Sweller, J., and Clark, R. E. (2006). Why minimal guidance during instruction does not work: an analysis of the failure of constructivist, discovery, problem-based, experiential, and inquiry-based teaching. Educational Psychologist. 41(2) 75-86. doi: 10.1207/s15326985ep4102_1
Koedinger, K. R., & Aleven, V. (2007). Exploring the Assistance Dilemma in Experiments with Cognitive Tutors. Educational Psychology Review 19(3) 239–264 doi: 10.1007/s10648-007-9049-0
Lipps, A., Trafton, J., & Gray, W. (1998). Animation as documentation: A Replication with Reinterpretation. In the Proceedings for the 45th annual conference of the Society for Technical Communication Arlington, VA. Retrieved from http://www.stc.org/proceedings/ConfProceed/1998/PDFs/00006.PDF
Mardia, K.V. (1970). Measures of multivariate skewness and kurtosis with applications. Biometrika, 57(3) 519-530. doi:10.1093/biomet/57.3.519
Mayer, R.E. (1997). Multimedia Learning: Are We Asking the Right Questions? Educational Psychologist, 32(1) 1-19. doi:10.1207/s15326985ep3201_1
Mayer, R. E. (2001). Multimedia Learning. Cambridge: Cambridge University Press
Mayer, R.E. (2004). Should there be a three-strikes rule against pure discovery learning? The case for guided methods of instruction. American Psychologist. 59(1) 14-19. doi: 10.1037/0003-066X.59.1.14
Mayer, R. E. (2005). Cognitive theory of multimedia learning. In R. E. Mayer (Ed.), Cambridge handbook of multimedia learning (pp. 31–48). New York: Cambridge University Press.
Mayer, R. E. (2009). Multimedia learning (2nd ed). New York: Cambridge University Press.
Mayer, R. E., & Anderson, R.B. (1991). Animations need narrations: an experimental test of a dual-coding hypothesis. Journal of Educational Psychology, 83(4) 484-490. doi:10.1037/0022-0663.83.4.484
Mayer, R.E., Bove, W., Bryman, A., Mars, R. and Tapangco, L. (1996). When less is more: meaningful learning from visual and verbal summaries of science textbook lessons. Journal of Educational Psychology, 88(1) 64-73. doi:10.1037/0022-0663.88.1.64
Mayer, R.E., & Chandler, P. (2001). When learning is just a click away: does simple user interaction foster deeper understanding of multimedia messages? Journal of Educational Psychology, 93 (2) 390-397. doi:10.1037/0022-0663.93.2.390
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 19
Mayer, R.E., Deleeuw, K.E., & Ayers, P. (2007). Creating retroactive and proactive interference in multimedia learning. Applied Cognitive Psychology, 21 795–809. doi: 10.1002/acp.1350
Mayer, R. E., & Moreno, R. (1998, April). A Cognitive Theory of Multimedia Learning: Implications for Design Principles . Paper presented at the annual meeting of the ACM SIGCHI Conference on Human Factors in Computing Systems, Los Angeles, CA. Retrieved from http://www.unm.edu/~moreno/PDFS/chi.pdf
Mayer, R.E., & Moreno, R. (2002). Aids to computer-based multimedia learning. Learning and Instruction, 12(1) 107–119. doi:10.1016/S0959-4752(01)00018-4
Mayer, R. E., & Sims, V. K. (1994). For whom is a picture worth a thousand words? Extensions of a dual-coding theory of multimedia learning. Journal of Educational Psychology, 84, 389–401. doi:10.1037/0022-0663.86.3.389
Merrill, M.D. (1983). Component Display Theory. In C. Reigeluth (Ed.), Instructional Design Theories and Models. Hillsdale, NJ: Erlbaum Associates.
Microsoft Excel 2003. (Version 11)[Computer program]. Redmond, WA: Microsoft
Moreno, R., & Mayer, R. E. (1999). Multimedia-supported metaphors for meaning making in mathematics. Cognition and Instruction, 17(3) 215–248. doi:10.1207/S1532690XCI1703_1
Mousavi, S.Y., Low, R. & Sweller, J. (1995). Reducing cognitive load by mixing auditory and visual presentation modes. Journal of Educational Psychology, 87(2) 319-334. doi:10.1037/0022-0663.87.2.319
Najjar, L. J. (1996). The effects of multimedia and elaborative encoding on learning (GIT-GVU-96-05). Retreived from Georgia Institute of Technology, Graphics, Visualization - Usability Center. http://hdl.handle.net/1853/3491
Paas, F. G. W. C. (1992). Training strategies for attaining transfer of problem-solving skill in statistics: A cognitive load approach. Journal Educational Psychology, 84 (4), 429-434. doi:10.1037/0022-0663.84.4.429
Paas, F. G. W. C., & van Merrienboer, J. J. G. (1993). The efficiency of instructional conditions: An approach to combine mental-effort and performance measures. Human Factors, 35(4), 737-743.
Paas, F., Tuovinen, J. E., Tabbers, H. K., & Van Gerven, P. W. M. (2003). Cognitive load measurement as a means to advance cognitive load theory. Educational Psychologist, 38 (1), 63–71. doi:10.1207/S15326985EP3801_8
Palmiter, S. (1991). Animated Demonstrations for learning procedural, computer-based tasks. (Doctoral Dissertation). Available from ProQuest Dissertation and Theses database. (UMI No. 9124076)
Palmiter, S. (1993). The effectiveness of animated demonstrations for computer-based tasks: a summary, model and future research. Journal of Visual Languages and Computing, 4 71-89. doi:10.1006/jvlc.1993.1005
Palmiter, S., & Elkerton, J. (1991). An evaluation of animated demonstrations of learning computer-based tasks. In the Proceedings of the SIGCHI conference on Human factors in computing systems: Reaching through technology. New Orleans, Louisiana, USA. p. 257 – 263. doi:10.1145/108844.108906
Penney, C.G. (1989). Modality effects and the structure of short-term memory. Memory and Cognition 17(4) 398-442. Retrieved from http://www.psychonomic.org/backissues/7537/mc/Issue%204/mc-17-398.pdf
Reimann, P. & Neubert, C. (2000). The role of self-explanation in learning to use a spreadsheet through examples. Journal of Computer Assisted Learning, 16(4), 316-325. doi:10.1046/j.1365-2729.2000.00144.x
Rourke, A., & Sweller, J. (2009). The worked example effect using ill-defined problems: learning to recognise designers' styles. Learning and Instruction, 19(2) 185–199. doi:10.1016/j.learninstruc.2008.03.006
SAS (2007). Macro to test multivariate normality. [Computer program]. Retrieved from http://support.sas.com/kb/24/983.html
Schwonke, R., Renkl, A., Krieg C., Wittwer, J., Aleven, V., & Salden, R. (2009). The worked example effect: Not an artefact of lousy control conditions. Computers in Human Behavior, 25(2) 258–266. doi:10.1016/j.chb.2008.12.011
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 20
Scoville, W.B., & Milner, B. (1957). Loss of recent memory after bilateral hippocampal lesions. Journal of Neurology, Neurosurgery, and Psychiatry, 20, 11-21. doi:10.1136/jnnp.20.1.11
Skinner, B. F. (1958). Teaching machines: From the experimental study of learning come devices which arrange optimal conditions for self-instruction. Science, 128 (3330) 969-977. doi:10.1126/science.128.3330.969
Squire, L. R., & Zola, S. M. (1996). Structure and function of declarative and nondeclarative memory systems. Proceedings of the National Academy of Sciences, 93, 13515-13522.
Squire, L.R. (1992). Memory and the hippocampus: a synthesis from findings with rats, monkeys, and humans. Psychological Review, 99 (2) 195-231. doi:10.1073/pnas.93.24.13515
Squire, L.R. (2008). Introduction to Memory Section. In Pomerantz, J.R. (Ed.) Topics In Integrative neuroscience: from cells to cognition. (pp. 239-242) Cambridge: Cambridge University Press
Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257-285. doi:10.1207/s15516709cog1202_4
Sweller, J. (2006). The worked-example effect and human cognition. Learning and Instruction, 16(2) 165-169. doi:10.1016/j.learninstruc.2006.02.005
Sweller, J., & Chandler, P. (1991). Evidence for cognitive load theory. Cognition and Instruction, 8(4) 351-362. doi:10.1207/s1532690xci0804_5
Sweller, J., & Cooper, G. A. (1985). The use of worked examples as a substitute for problem solving in learning algebra. Cognition and Instruction, 2(1), 59-89. doi:10.1207/s1532690xci0201_3
TechSmith Morae (Version 1.0.1) [Computer program] Okemos, MI: TechSmith
TechSmith Camtasia Studio (Version 4.0) [Computer program] Okemos, MI: TechSmith
Tuovinen, J. E., & Sweller, J. (1999). A comparison of cognitive load associated with discovery learning and worked examples. Journal of Educational Psychology, 91(2), 334-341. doi:10.1037/0022-0663.91.2.334
Tuovinen, J. E., & Paas, F. G. W. C. (2004). Exploring multidimensional approaches to the efficiency of instructional conditions. Instructional Science, 32(1-2) 133–152. doi:10.1023/B:TRUC.0000021813.24669.62
Tuovinen, J. E., & Sweller, J. (1999). A comparison of cognitive load associated with discovery learning and worked examples. Journal of Educational Psychology, 91(2), 334-341. doi:10.1037/0022-0663.91.2.334
Tversky, B., Morrison, J. B., & Betancourt, M. (2002). Animation: Can it facilitate? International Journal of Human–Computer Studies, 57, 247–262. doi:10.1006/ijhc.2002.1017
Ward, M., & Sweller, J. (1990). Structuring effective worked examples. Cognition and Instruction, 7(1), 1-39. doi:10.1207/s1532690xci0701_1
Waterson, P. & O’Malley, C. (1993). Using animated demonstrations in multimedia applications: Some suggestions based upon experimental evidence. In G. Salvendy & M. Smith (Eds.), Human-Computer Interaction: Software and Hardware Interfaces, Proceedings of the 5th International Conference on Human-Computer Interaction (August 8-13, Orlando, Florida) Vol. 2 (p. 543-548). Amsterdam: Elsevier.
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 21
Figure 1. Week 1 Relative Condition Efficiency (RCE1)
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 22
Figure 2. Week 1 Performance Efficiency (PE1)
Running head: AN ANALYSIS OF THE WORKED-EXAMPLE EFFECT 23
Table 1
Picnic problem accuracy rubric
flip layer rotate move item *** umbrella *** tshirt
*** *** head *** *** *** right leg *** *** head 2 *** *** purple shirt *** hat *** *** *** s left leg *** *** *** bent right leg *** *** *** left leg *** green shorts *** *** *** arm 2 *** pink shorts *** *** *** left arm *** *** body *** *** picnic basket *** *** *** arm *** *** right arm *** *** torso *** *** table *** *** bird3 *** *** *** bird2 *** *** bird1
0 0 0 0 0